Oracle and Google Cloud Enable Bidirectional Iceberg Writes Across Database and Analytics

Oracle and Google Cloud eliminate ETL between database and lakehouse

Oracle AI Database@Google Cloud can now query and write to Apache Iceberg tables in Google Cloud Storage without replicating data or running ETL pipelines. The preview feature allows Oracle Autonomous AI Database workloads to perform DML operations against shared Iceberg tables accessible to BigQuery and Managed Spark, across public cloud and on-premises environments.

The development addresses a specific source of multi-cloud waste: enterprises maintaining duplicate copies of data because their transactional databases and analytics platforms cannot share the same storage layer. By using Iceberg as a common table format, Oracle customers can run analytics queries in Google Cloud against live database tables without exporting, copying, or scheduling batch transfers.

The announcement disclosed no pricing model, throughput benchmarks, customer count, or quantified cost reduction. Oracle and Google Cloud positioned the integration as bidirectional — analytics tools can read database tables, and database workloads can write back to lakehouse storage — but provided no detail on write consistency, transaction semantics, Iceberg catalog compatibility, or regional availability.

This places Oracle and Google Cloud against AWS–Snowflake, Databricks, Microsoft Fabric, and native cross-cloud lakehouse products. The competitive claim is direct interoperability between Oracle's database stack and Google's analytics services rather than requiring a third platform or replication layer. Buyers with Oracle databases and Google Cloud analytics should evaluate whether eliminating ETL pipelines offsets cross-cloud egress charges, Oracle database licensing costs, and the operational risk of running a preview feature in production.

Lumen and Microsoft expand Azure co-selling, disclose no commercial terms

Lumen Technologies and Microsoft announced an expanded collaboration to co-sell Lumen's cloud networking and application connectivity portfolio through Azure's ecosystem. Lumen with Alkira will serve as a portal for connecting, securing, and managing hybrid and multi-cloud networks, with Lumen positioning itself as an Azure-focused networking partner.

The November 2 announcement contained no contract value, customer count, subscription pricing, deployment scale, latency improvement, or service-level commitment. It describes a sales relationship and planned availability rather than a product launch with measurable outcomes.

Large enterprises may gain a simpler procurement path for multi-cloud connectivity and centralized policy management, potentially consolidating networking vendors on an RFP. The strategic value depends on whether Lumen's portfolio becomes commercially available through Azure Marketplace or equivalent channels, which clouds are supported, how identity and policy integrate with Azure Entra, and whether total cost undercuts native Azure Virtual WAN, AWS Transit Gateway, or competing SASE and interconnect products from Cisco, Aviatrix, Equinix, and Megaport.

Buyers should treat this as a channel development signal, not evidence of reduced networking cost or operational complexity. The absence of pricing and benchmarks prevents a business-case comparison.

Snowflake positions Cortex AI Gateway as cross-model governance layer

Snowflake promoted Cortex AI Gateway as a control plane for governing AI agent access to models and enterprise data across multiple clouds. The company said Glean and Kargo are using the gateway, while Uniphore is extending its controls to agents it orchestrates for joint customers.

The announcement identified at least two named users and one named partner but provided no customer totals, pricing, request-volume data, latency results, or measured cost reduction. The competitive issue is shifting from hosting models to governing model calls, agent identities, data access, and spend across AWS Bedrock, Azure OpenAI, Google Vertex AI, and independent providers.

Enterprises designing multi-model or multi-cloud agent architectures should evaluate whether a centralized gateway reduces duplicated policy controls and improves chargeback visibility. The primary risk is platform dependence: a gateway may simplify governance while deepening reliance on Snowflake's security, data catalog, and AI control plane. Buyers should require usage metrics, egress cost analysis, policy enforcement audits, and total cost comparisons against hyperscaler-native governance services before committing budget.

What to watch

Data portability is becoming more actionable. Oracle and Google Cloud's Iceberg integration addresses a concrete multi-cloud problem — copying data between transactional and analytics environments — but the lack of pricing and performance data limits near-term confidence. Buyers should treat this as a candidate for a controlled pilot, not a production-ready business case.

Networking partnerships remain light on buyer-grade evidence. Lumen and Microsoft's expanded collaboration may improve Azure-centric procurement, but the announcement provides no cost, latency, or operational improvement to justify prioritization over incumbent networking vendors.

AI governance is emerging as the next multi-cloud control point. Snowflake's gateway push indicates enterprises may buy a policy and observability layer above multiple model providers, competing with native services from AWS, Azure, and Google Cloud. Named deployments make this more concrete than generic interoperability claims, but ROI comparisons require usage, egress cost, and policy enforcement metrics that the announcement does not provide.